The index rainfall in the regional frequency analysis of extreme events in Tuscany (Italy)

Abstract

AbstractIn the framework of regional frequency analysis of annual maximum of rainfall depth for short duration storms, the aim of this work is to find expressions of the scale factor of a non‐dimensional regional probability distribution, that is the index rainfall. Over a territory of about 22 000 km2in Tuscany (Italy), the time series of precipitation extreme events of 470 rain gauges, in the period from 1923 to 2002, have been collected and investigated. The Tuscany Region has been divided into three zones and sub‐zones on the basis of a hierarchical regionalization procedure, based on the index rainfall method, and the two‐component extreme value (TCEV) distribution, as the probabilistic model. On the time series, characterized by a low spatial correlation, a local extreme value analysis, based on the generalized extreme value (GEV) distribution, has been also achieved. At the gauge site, the authors introduce the hypotheses that the estimated GEV corresponds to the estimated extreme value of the regional frequency analysis. The index rainfall at the gauge site can be therefore assumed as the ratio of the GEV distribution estimated rainfall and the probabilistic growth factor defined in the hierarchical regionalization procedure. On the basis of the assumption above, a multivariate analysis has been also proposed, with the aim to model the relationship among the spatial variability of index rainfall and the territory characteristics. Copyright © 2008 John Wiley & Sons, Ltd.